Analysis of large particle sizes using a machine vision system

Abstract

Many methods based on machine vision were used to estimate coarse particles size distribution in recent years, but comparison of accuracy parameters representing particle size has not been carried out and a related representing analysis has not been yet proposed. Nine parameters were investigated. The results indicated the minor axis of equivalent ellipse and breadth of the best-fit rectangle were the most suitable for representing particle size. The former accuracy ratio was 86.43% and the latter accuracy ratio was 85.39%, while the accuracy of other parameters was less than 70%. A related representing analysis was proposed to explain this phenomenon. This research is instructive and meaningful for the size distri- bution estimation by machine vision.